Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Gracefully CUDF_FAIL when skip_rows > 0 in Chunked Parquet reader #16385

Merged
merged 1 commit into from
Jul 24, 2024
Merged

Gracefully CUDF_FAIL when skip_rows > 0 in Chunked Parquet reader #16385

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions cpp/src/io/parquet/reader.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -41,6 +41,11 @@ chunked_reader::chunked_reader(std::size_t chunk_read_limit,
rmm::cuda_stream_view stream,
rmm::device_async_resource_ref mr)
{
// TODO: skip_rows not currently supported in chunked parquet reader until
// https://github.com/rapidsai/cudf/issues/16186 is closed
CUDF_EXPECTS(options.get_skip_rows() == 0,
"skip_rows > 0 is not currently supported in the Chunked Parquet reader.");

_impl = std::make_unique<impl>(
chunk_read_limit, pass_read_limit, std::move(sources), options, stream, mr);
}
Expand Down
29 changes: 19 additions & 10 deletions cpp/tests/io/parquet_chunked_reader_test.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1544,7 +1544,8 @@ TEST_F(ParquetChunkedReaderTest, TestNumRowsPerSource)

// Chunked-read rows_to_read rows skipping rows_to_skip from single data source
{
auto const rows_to_skip = 1'237;
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Were these tests segfaulting before this change?

Copy link
Member Author

@mhaseeb123 mhaseeb123 Jul 24, 2024 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Surprisingly no. The segfaults particularly show up if we have at least one string column in which case the chunking across subpass and output chunks goes haywire. (We certainly do have arithmetic errors in those calculations). Changed this as we will certainly get a CUDF_FAIL() after this PR.

// TODO: rows_to_skip = 0 until https://github.com/rapidsai/cudf/issues/16186 is resolved
auto const rows_to_skip = 0; // 1'237
auto const rows_to_read = 7'232;
auto constexpr output_read_limit = 1'500;
auto constexpr pass_read_limit = 3'500;
Expand All @@ -1571,7 +1572,8 @@ TEST_F(ParquetChunkedReaderTest, TestNumRowsPerSource)

// Chunked-read two data sources skipping the first entire file completely
{
auto constexpr rows_to_skip = 15'723;
// TODO: rows_to_skip = 0 until https://github.com/rapidsai/cudf/issues/16186 is resolved
auto constexpr rows_to_skip = 0; // 15'723;
auto constexpr output_read_limit = 1'024'000;
auto constexpr pass_read_limit = 1'024'000;

Expand All @@ -1588,20 +1590,25 @@ TEST_F(ParquetChunkedReaderTest, TestNumRowsPerSource)

auto const [result, num_chunks, num_rows_per_source] = read_table_and_nrows_per_source(reader);

// TODO: Enable code inside /* */ when https://github.com/rapidsai/cudf/issues/16186 is resolved
auto int64_col_selected =
int64s_col(int64_data.begin() + rows_to_skip - num_rows, int64_data.end()).release();
int64s_col(int64_data.begin() /* + rows_to_skip - num_rows */, int64_data.end()).release();

cudf::table_view const expected_selected({int64_col_selected->view()});

CUDF_TEST_EXPECT_TABLES_EQUAL(expected_selected, result->view());
// TODO: Enable the following check when https://github.com/rapidsai/cudf/issues/16186
// is resolved
// CUDF_TEST_EXPECT_TABLES_EQUAL(expected_selected, result->view());

EXPECT_EQ(num_rows_per_source.size(), 2);
EXPECT_EQ(num_rows_per_source[0], 0);
EXPECT_EQ(num_rows_per_source[1], nsources * num_rows - rows_to_skip);
EXPECT_EQ(num_rows_per_source[0], num_rows /* 0 */);
EXPECT_EQ(num_rows_per_source[1], num_rows /* nsources * num_rows - rows_to_skip */);
}

// Chunked-read from single data source skipping rows_to_skip
{
auto const rows_to_skip = 1'237;
// TODO: rows_to_skip = 0 until https://github.com/rapidsai/cudf/issues/16186 is resolved
auto const rows_to_skip = 0; // 1'237;
auto constexpr output_read_limit = 1'500;
auto constexpr pass_read_limit = 1'800;

Expand Down Expand Up @@ -1736,7 +1743,8 @@ TEST_F(ParquetChunkedReaderTest, TestNumRowsPerSourceMultipleSources)

// Chunked-read rows_to_read rows skipping rows_to_skip from eight data sources
{
auto const rows_to_skip = 25'571;
// TODO: rows_to_skip = 0 until https://github.com/rapidsai/cudf/issues/16186 is resolved
auto const rows_to_skip = 0; // 25'571;
auto const rows_to_read = 41'232;
auto constexpr output_read_limit = 15'000;
auto constexpr pass_read_limit = 35'000;
Expand Down Expand Up @@ -1782,8 +1790,9 @@ TEST_F(ParquetChunkedReaderTest, TestNumRowsPerSourceMultipleSources)

// Chunked-read four data sources skipping three files completely
{
auto const nsources = 4;
int constexpr rows_to_skip = num_rows * 3 + 1;
auto const nsources = 4;
// TODO: rows_to_skip = 0 until https://github.com/rapidsai/cudf/issues/16186 is resolved
int constexpr rows_to_skip = 0; // num_rows * 3 + 1;
auto constexpr output_read_limit = 15'000;
auto constexpr pass_read_limit = 35'000;
std::vector<int64_t> int64_selected_data{};
Expand Down
Loading